Lung cancer is the leading cause of death from cancer. More people die each year of bronchial carcinoma than of colon, breast, and prostate cancer combined. More than 50% of the patients will have distant metastases at diagnosis and only 20-25% of these will be localized and potentially resectable, with a five-year survival of 14%. Prior chest radiographs and sputum cytology studies lead to clinically meaningful improvements in stage distribution, resectability and survival, but no disease-specific mortality reductions have been demonstrated. More recently, these techniques have evolved to those of screening by low-dose spiral computed tomography and by the use of specific biomarkers for early detection, thus bringing back interest in lung cancer screening. The authors review screening for lung cancer made in the past, analyze more recent prospective studies, and the prospects for the future, and they point to their main limitations, methodological problems in design, major biases, all of which may invalidate the interpretation of results.